Research on Single-Trial ErrP Decoding Algorithm and Cognitive Enhancement Paradigm Design
Chang Niu, Zhuang Yan, Kuiying Yin, Shenghua Zhou · 2024
The Error-Related Potential (ErrP) is is a subtle indicator of erroneous behavior. This paper discusses the application value of ErrP and proposes a cognitive enhancement paradigm, using a visual stimulation paradigm that combines steady-state stimuli with rapid visual presentation. Data collection experiments in this paper were conducted based on this paradigm, and a study on single-trial ErrP classification algorithms was carried out. Improvements were made to the Discriminative Canonical Pattern Matching (DCPM) algorithm. Comparison of the participants' behavioral data and EEG classification results indicates that Gamma rhythmic background stimuli positively modulate the participants' discriminative cognitive processes. Under this paradigm, the participants' judgment efficiency and performance of the classification algorithm were significantly better than those under traditional paradigms, thus effectively improving the quality and data utilization rate of ErrP data collection. Meanwhile, the improved DCPM proposed in this paper shows better decoding capability for ErrP components compared to the original DCPM and other mainstream event-related potential decoding algorithms. In the dataset collected from 20 participants, the decoding balance accuracy increased by 3.7% to 79.6%, and the true positive rate increased by 8.2% to 87.5%.